The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes standard command-line tools such as git (status, log, diff, commit) and grep to perform its tasks, such as verifying implementation consistency and managing version control for documents. These operations are conducted within the local repository context.
[PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection by processing external project files. Ingestion points: {slug}-design.md, {slug}-checklist.yaml, and project source code (via grep and git diff). Boundary markers: The skill uses structured Markdown headers for reports but does not explicitly delimit untrusted file content. Capability inventory: The skill can write to the local filesystem (acceptance reports, architecture docs), execute tests, and perform git commit. Sanitization: The risk is mitigated by mandatory human review checkpoints and explicit user confirmation required before any permanent changes or commits are finalized.
[SAFE]: No external network communication, obfuscation, or unauthorized access to sensitive system files (e.g., credentials) was detected. All operations are confined to the intended development environment.

easysdd-feature-acceptance